PRACTICAL COMPUTATION OF STATISTICAL PROPERTIES OF RAINFLOW COUNTS

Some mathematical properties of the rainflow counting method were esta blished by Rychlik. They enabled him to propose a new algorithm for ra inflow counting. Implementations of this algorithm for signal analysis would show poor speed performance, but it is of utmost interest as it can be used abstractly on the statistical properties of the input seq uence. The practical implementation of such derivation is discussed he re, and probability density functions (PDFs) for rainflow cycles betwe en level j and level k are derived in the simple case when peaks and t roughs are chosen from known PDFs in a completely random succession. A lgorithms are provided in a common practice case when the sequence is described as a Markovian process with given transition probability mat rix. Sequences for which the statistical properties are inferred from the power spectral densities are also discussed. Application is made t o some standard sequences from the offshore and automotive industry.